Update for json-schema-org#167 #2
Conversation
| In addition to a boolean value or an object using schema kewyords defined | ||
| in the meta-schema, a schema may be represnted by an object containing a "$ref" property. | ||
| The value of the $ref is a URI Reference. | ||
| The "$ref" keyword is used to reference an externally defined schema, and provides the ability to validate recursive structures through self-reference. |
There was a problem hiding this comment.
"externally" suggests "external to the JSON Schema document" to me, but I primarily use "$ref" within the same document. What meaning were you going for here?
There was a problem hiding this comment.
The intention was to indicate schemas don't necessarily have to be inside the current one; it wasn't intended to mean exclusively. I'll take this out.
| Resolved against the current URI base, it identifies the URI of a schema to use. | ||
| The referenced schema MUST NOT itself contain a "$ref" property. |
There was a problem hiding this comment.
I disagree with this. While a double (or worse) reference is somewhere between silly and a bad idea, it could easily come up when different people maintain different parts of an ecosystem.
This is the kind of thing that would be caught by a linter rather than spec enforcement. I could support a SHOULD NOT here, though. Would that work for you?
There was a problem hiding this comment.
The issue the $ref behavior is only guaranteed for subschemas, this is one way of enforcing that.
The behavior can be done by placing it inside an "allOf" if desired.
But I suppose this is an issue for a... new issue anyways.
There was a problem hiding this comment.
I'm not sure I follow here- when we spoke on IRC the idea was that both booleans and references would be valid schemas anywhere. It's a bit weird to have a schema that's just a reference but there could be a reason for it. I still think SHOULD NOT is more correct, but I think I might be missing part of your point.
|
@awwright not sure how email notifications work when I comment on an "outdated" diff, so putting it here too: I'm not sure I follow the "only subschemas" line of thought here- when we spoke on IRC the idea was that both booleans and references would be valid schemas anywhere. It's a bit weird to have a schema that's just a reference but there could be a reason for it. I still think SHOULD NOT is more correct, but perhaps I'm just missing part of your point? |
|
@awwright Since I have filed json-schema-org#174 for the question of a reference being a valid root schema, I will go ahead and merge this in. We can modify it if we resolve json-schema-org#174 in favor of allow references as root schemas. If it goes the other way, this is a good clarification so might as well get it in now. |
Some edits for json-schema-org#167
subschema is either an object or a booleanline informative instead of normative (it's a fact that's already true of all schemas), and